Drinking rapidly abolishes thirst and vasopressin secretion in dehydrated humans before major changes in plasma osmolality are observed. We studied the effects of drinking on plasma vasopressin and thirst in seven healthy volunteers rendered hypernatremic by the infusion of hypertonic (855 mmol/l) sodium chloride solution. Thirst was measured on a visual analogue scale (0-10 cm). Infusion of hypertonic saline caused linear increases in plasma osmolality (289 +/- 1 to 306 +/- 1 mosmol/kg, mean +/- SE, P less than 0.001), plasma vasopressin (0.6 +/- 0.2 to 6.4 +/- 1.9 pmol/l, P less than 0.001), and thirst (1.4 +/- 0.4 to 7.4 +/- 0.5 cm, P less than 0.001). Water was allowed 15 min after cessation of the infusion, and within 5 min of drinking both plasma vasopressin and thirst were significantly lower than postinfusion. After 20 min of drinking, plasma vasopressin had fallen from 6.5 +/- 0.9 to 1.3 +/- 0.3 pmol/l (P less than 0.001) and thirst from 7.7 +/- 0.5 to 1.0 +/- 0.2 cm (P less than 0.001) despite no significant change in plasma osmolality (306 +/- 0.9 to 304 +/- 0.8 mosmol/kg, P = 0.17), and the drinking of 1,200 +/- 60 ml of water, over 85% of the mean cumulative water intake in the 30-min drinking period. Control studies in the same subjects showed comparable rises in plasma vasopressin, plasma osmolality, and thirst during hypertonic saline infusion but no fall in any of these parameters during an equivalent 30-min period after the infusions, during which water was withheld.(ABSTRACT TRUNCATED AT 250 WORDS)